No, you can’t. You need to enable flexible parallelism (FP) before you create the database.
There are limited benefits for some queries if FP is not turned on when you create(d) the database, but in general FP is not enabled. E.g. older databases can do some parallel scan operations at the segment granule size, using more threads than there are partitions, but shard-key-aware operations will not benefit from flexible parallelism. For example, colocated joins cant use FP for non-FP databases.